close

Вход

Забыли?

вход по аккаунту

?

j.ejps.2017.10.022

код для вставкиСкачать
Accepted Manuscript
Coupled in silico platform: Computational fluid dynamics (CFD)
and physiologically-based pharmacokinetic (PBPK) modelling
Aleksandra Vulovic, Tijana Sustersic, Sandra Cvijic, Svetlana
Ibric, Nenad Filipovic
PII:
DOI:
Reference:
S0928-0987(17)30577-8
doi:10.1016/j.ejps.2017.10.022
PHASCI 4268
To appear in:
European Journal of Pharmaceutical Sciences
Received date:
Revised date:
Accepted date:
1 June 2017
11 October 2017
14 October 2017
Please cite this article as: Aleksandra Vulovic, Tijana Sustersic, Sandra Cvijic, Svetlana
Ibric, Nenad Filipovic , Coupled in silico platform: Computational fluid dynamics (CFD)
and physiologically-based pharmacokinetic (PBPK) modelling. The address for the
corresponding author was captured as affiliation for all authors. Please check if
appropriate. Phasci(2017), doi:10.1016/j.ejps.2017.10.022
This is a PDF file of an unedited manuscript that has been accepted for publication. As
a service to our customers we are providing this early version of the manuscript. The
manuscript will undergo copyediting, typesetting, and review of the resulting proof before
it is published in its final form. Please note that during the production process errors may
be discovered which could affect the content, and all legal disclaimers that apply to the
journal pertain.
ACCEPTED MANUSCRIPT
Coupled in silico platform: Computational Fluid Dynamics (CFD) and Physiologically-Based
Pharmacokinetic (PBPK) modelling
Authors: Aleksandra Vulovica,b, Tijana Sustersica,b, Sandra Cvijicc, Svetlana Ibricc, Nenad
Faculty of Engineering, University of Kragujevac, Sestre Janjić 6, Kragujevac, Serbia
b
CR
a
IP
T
Filipovica,b
BioIRC, Bioengineering Research and Development Center, Prvoslava Stojanovica 6,
Department of Pharmaceutical Technology and Cosmetology, University of Belgrade-Faculty of
AN
c
US
Kragujevac, Serbia, aleksandra.vulovic@kg.ac.rs, tijanas@kg.ac.rs, fica@kg.ac.rs
Pharmacy, Vojvode Stepe 450, Belgrade, Serbia, gsandra@pharmacy.bg.ac.rs,
ED
M
ibric@pharmacy.bg.ac.rs
Corresponding author:
PT
Sandra Cvijic
CE
Department of Pharmaceutical Technology and Cosmetology
University of Belgrade - Faculty of Pharmacy
AC
Vojvode Stepe 450
11221 Belgrade, Serbia
Tel: +381638611341
Fax: +381113972840
E-mail address: gsandra@pharmacy.bg.ac.rs
1
ACCEPTED MANUSCRIPT
Abstract
One of the critical components of the respiratory drug delivery is the manner in which the
inhaled aerosol is deposited in respiratory tract compartments. Depending on formulation
properties, device characteristics and breathing pattern, only a certain fraction of the dose will
T
reach the target site in the lungs, while the rest of the drug will deposit in the inhalation device or
IP
in the mouth–throat region. The aim of this study was to link the Computational Fluid Dynamics
CR
(CFD) with Physiologically-Based Pharmacokinetic (PBPK) modelling in order to predict
aerolisolization of different dry powder formulations, and estimate concomitant in vivo
US
deposition and absorption of amiloride hydrochloride. Drug physicochemical properties were
AN
experimentally determined and used as inputs for the CFD simulations of particle flow in the
generated 3D geometric model of Aerolizer® dry powder inhaler (DPI). CFD simulations were
M
used to simulate air flow through Aerolizer® inhaler and Discrete Phase Method (DPM) was used
ED
to simulate aerosol particles deposition within the fluid domain. The simulated values for the
percent emitted dose were comparable to the values obtained using Andersen cascade impactor
PT
(ACI). However, CFD predictions indicated that aerosolized DPI have smaller particle size and
CE
narrower size distribution than assumed based on ACI measurements. Comparison with the
literature in vivo data revealed that the constructed drug-specific PBPK model was able to
AC
capture amiloride absorption pattern following oral and inhalation administration. The PBPK
simulation results, based on the CFD generated particle distribution data as input, illustrated the
influence of formulation properties on the expected drug plasma concentration profiles. The
model also predicted the influence of potential changes in physiological parameters on the extent
of inhaled amiloride absorption. Overall, this study demonstrated the potential of the combined
2
ACCEPTED MANUSCRIPT
CFD-PBPK approach to model inhaled drug bioperformance, and suggested that CFD generated
results might serve as input for the prediction of drug deposition pattern in vivo.
Keywords: computational fluid dynamics; discrete phase modelling; physiologically-based
AC
CE
PT
ED
M
AN
US
CR
IP
T
pharmacokinetic modelling; amiloride; dry powder inhaler; particle size distribution
3
ACCEPTED MANUSCRIPT
1. Introduction
Formulation of inhaled medicines is rather challenging, as a number of factors influence quality
and efficacy of the final product. Changes in physiological parameters between individuals or
population groups further impede prediction of drug product bioperformance based on drug and
T
formulation characteristics. In order to accelerate the development of safer and more effective
IP
patient-tailored inhaled medicines and reduce the number of in vivo studies, formulation science
CR
needs to embrace new methods for the evaluation of drugs/dosage forms, and focus on modelbased development of inhaled products. In this context, in silico predictive tools can facilitate
US
rational design of inhaled pharmaceutics, which is in accordance with the Quality by Design
AN
(QbD) paradigm in pharmaceutical development, incorporated in ICH Q8(R2) guidance (ICH,
2005). Computational models and advanced simulation tools can upgrade our understanding of
M
the complex physiological processes, and interactions a drug goes after inhalation, and thus
ED
facilitate the development of inhalation medicines.
PT
1.1. Computational fluid dynamic (CFD) Analysis and Discrete Phase Modelling (DPM)
CE
Technology development, such as computer power becoming cheaper and available software
more powerful, led to increased use of computer-based simulation in different branches of
AC
science and engineering (Wong et al., 2012).
Dry powder inhalers (DPIs) deliver drugs in the form of a dry powder. Aerosol particles that are
generated by this type of inhalers show good results and offer advantages such as fast delivery of
particles to the lungs (Clark, 2004; Chan, 2006). DPIs do not require use of propellants, they lead
to higher deposition of aerosol particles within the lungs and also simplify the inhalation
technique. DPIs help to reduce the fluctuation of the inhaled dose compared to other available
4
ACCEPTED MANUSCRIPT
inhalers such as Metered Dose Inhalers (Brocklebank et al., 2001; Terzano, 2008) as well as
reduce systemic side effects compared to other drug delivery routes (Hoppentocht et al., 2014).
The problem with these devices is their inefficiency, since less than 30% of dose reaches the
lungs (Islam and Gladki, 2008). In order to improve the performance of these inhalers we need to
T
understand the dispersion and deposition mechanisms better.
IP
Computational fluid dynamics (CFD) can be used for modelling both laminar and turbulent flow
CR
(Wong et al., 2012). The disadvantages of the CFD approach are that it can only simulate airflow
and it is not possible to consider particle interaction. The advancement in the field of CFD
US
analysis has created a possibility for a new approach for inhalator modelling. This advancement
AN
allowed for aerosol transport and deposition to be calculated in realistic three-dimensional (3-D)
models of the inhaler (Longest and Holbrook, 2012).
M
Several computational fluid-particle dynamics (CF-PD) models are available for the calculation
ED
of air-drug mixture dynamics (Feng and Kleinstreuer, 2014), such as the discrete phase model
(DPM), two-fluid model, mixture model, dense dispersed phase model (DDPM), and the discrete
PT
element method (DEM) (Kleinstreuer et al., 2014). In order to gain knowledge on particle
CE
dispersion we need to couple CFD with DPM. Some authors investigated commercial DPIs using
CFD-DPM analysis in order to gain better knowledge on dispersion mechanisms. Zhou at al.
AC
(2013) researched the effect of device design of the commercial DPI, the Aerolizer® on the
aerosolization of a carrier-based DPI formulation (Foradile®) using a combination of CFD and
DPM, while Jiang et al. (2012) using the same method investigated the effect of powder
residence time on the performance of the Aerolizer DPI. CFD coupled with experimental powder
dispersion analysis, can provide an initial quantification of the turbulence levels and average
5
ACCEPTED MANUSCRIPT
particle impaction velocities. This information can help maximize the dispersion performance of
a DPI (Coates et al., 2006).
The findings presented by Mossaad (2014) suggest that the deposition profile of inhalation dry
powders is affected by two major independent factors:
T
(1) patient-related factors; which can be cited as the anatomical and physiological aspects of the
IP
respiratory system as well as the inhalation airflow rate;
CR
(2) physical properties of dry powder formulations; which include properties of pure drug as well
as properties of nanocarrier systems
US
Most marketed DPIs are passive inhalers that employ the patient's inspiratory effort to generate
AN
the necessary airflow, and the associated turbulence, to overcome the cohesive nature of the
respirable active pharmaceutical ingredient (API) and fluidize the powder bed into a respirable
ED
has on the in vitro DPI performance.
M
aerosol (Wong et al., 2012). Voss and Finlay (2002) investigated the effect that air turbulence
Coates and co-workers have analyzed the effect of the Aerolizer® design on device performance.
PT
They have analyzed the influence of airflow (Coates et al., 2005a), grid structure and mouthpiece
CE
length (Coates et al., 2004), mouthpiece geometry (Coates et al., 2007), inlet size (Coates et al.,
2006) as well as the role of the capsule (Coates et al., 2005b). They have found that:
increased flow rate through the Aerolizer® leads to increased level of turbulence (Coates
AC

et al., 2005a);

the grid straightens the airflow exiting the device, and increase in grid voidage leads to
the increase of the amount of mouthpiece retention (Coates et al., 2004);

the length of the mouthpiece has no significant effect on the performance of the device
(Coates et al., 2004);
6
ACCEPTED MANUSCRIPT

widening of the Aerolizer® exit makes a small difference to the dispersion performance,
and it reduces the axial velocity of the exiting airflow which leads to reduced throat
impaction (Coates et al., 2007);

the air inlet size reduction leads to increased turbulence and particle impaction velocities
T
within the device at low flow rates, which enhance the dispersion performance (Coates et
the capsule reduced the turbulence levels within the device, whilst its size does not have
CR

IP
al., 2006);
US
significant effect on performance (Coates et al., 2005b).
Information obtained using computer based simulation (CFD and DPM analysis) help us
AN
understand dispersion mechanisms better. This knowledge can be used for inhaler design
M
changes in order to create more efficient device.
ED
1.2. Physiologically-based pharmacokinetic (PBPK) modelling
PT
Physiologically-based pharmacokinetic (PBPK) modelling is a relatively new concept pointed
out to be beneficial for the in silico simulation of drug pharmacokinetic behavior following drug
CE
administration through different dosing sites. In comparison to simpler compartmental
pharmacokinetic models, which have no physiological meaning, PBPK models forecast drug
AC
pharmacokinetics as a function of physiology characteristics (Bouzom et al., 2012; Jones and
Rowland-Yeo, 2013). These properties may change in disease states or due to inter- or intrasubject variations, and if the relevant physiological properties are known, PBPK model can be
used to predict drug pharmacokinetics in target population or tailor drug dosing for the purpose
of personalized therapy (Suri et al., 2015, Almukainzi et al., 2016). In addition, PBPK models
can provide mechanistic explanation for unusual or unexpected data observed in vivo (Peters,
7
ACCEPTED MANUSCRIPT
2008; Samanth et al., 2017). Another advantage of PBPK models is that they can serve as a basis
for the prediction of drug-drug interactions (Zhou et al., 2016; Zhuang and Lu, 2016), which is of
particular importance for poly-medicated patients. There have also been attempts to link PBPK
models with drug pharmacodynamic properties, and predict drug exposure and
T
pharmacodynamic effect (Chetty et al., 2014, Alqahtani and Kaddoumi, 2016).
IP
Most efforts have been directed to the development of predictive PBPK models for orally
CR
administered drugs, while mechanistic PBPK models for other dosing routes are relatively new,
and still poorly exploited. There are several software tools for PBPK modelling of inhaled drug
US
absorption (Borghardt et al., 2015; Fröhlich et al., 2016). Among these is GastroPlusTM software
AN
package, with the integrated Pulmonary Compartmental Absorption & Transit (PCAT™) model
able to predict the absorption and disposition of drugs administered via intranasal and/or
M
respiratory route. Modelling and simulation with this software can be used to assess the influence
ED
of drug properties, drug dose and/or dosage form characteristics on drug in vivo
absorption/disposition, as well as to mechanistically interpret the influence of different
PT
physiological or disease related factors on drug absorption. However, published examples of the
2016).
CE
application of PCAT™ model are still scarce (Wu et al., 2013; Backman et al., 2016; Wu et al.,
AC
One of the critical factors affecting the reliability of the PBPK modelling results is the selection
of input data that characterize dosage form properties, such as size distribution of particles
leaving the inhaler. Common methods to assess aerodynamic properties of aerosolized particles
rely on in vitro tests using impingers or impactors (Mitchell and Nagel, 2003). These methods
are predominantly designed to provide data on lung particle deposition, and they are used for the
in vitro assessment of inhaled drugs bioequivalence. However, there are certain issues about the
8
ACCEPTED MANUSCRIPT
accuracy of these methods (de Boer et al., 2002), and the obtained results (expressed as mass
median aerodynamic diameter (MMAD), and geometric standard deviation (GSD)) might not
necessarily reflect actual aerosolized particle size distribution (PSD). As an alternative to the in
vitro determination of MMAD and GSD, CFD simulations may provide data on PSD, which can
T
further be used as input for the calculation of deposition fractions in different lung
IP
compartments.
CR
The purpose of this study was (i) to link the CFD simulations of particle distribution from DPI
with PBPK modelling in order to predict in vivo deposition and absorption of the inhaled model
US
drug, amiloride hydrochloride. The additional goals were (ii) to compare CFD simulation result
AN
with the in vitro data obtained using Andersen cascade impactor (ACI), and to assess possible
advantages of this in silico approach to accurately depict size distribution of aerosolized
M
particles; (iii) to mechanistically interpret the influence of formulation properties and
ED
physiological factors on amiloride deposition and absorption following inhalation.
Amiloride is a sodium channel blocker with weak diuretic properties, commonly used to treat
PT
oedema associated with hepatic cirrhosis and heart failure. Amiloride, as hydrochloride (HCl)
CE
salt, has also been used in the management of cystic fibrosis, due to its ability to reduce sodium
and water absorption from the lung luminal membrane and preserve airway surface liquid.
AC
Inhaled amiloride acts locally in the bronchiolar airway epithelium, but if absorbed in large
percent, high plasma concentrations may cause hyperkalemia and undesirable cardiovascular
effects (Jones et al, 1997). Therefore, it is important to understand the disposition and plasma
concentration-time profile following inhalation of this drug.
2. Materials and methods
9
ACCEPTED MANUSCRIPT
2.1. Drug physicochemical and aerodynamic properties
Drug-related input parameters required for CFD and PBPK simulations were experimental data
taken from literature or in silico estimated values. The selected data and appropriate references
are denoted in the relevant sections.
T
Physicochemical and aerodynamic properties of the model amiloride HCl DPI formulations were
IP
previously determined in vitro as described in the study of Djokic et al. (2014). Four jet-milled
CR
samples (JM1-JM4) were prepared under different experimental setups, with varying three jetmilling process variables (i.e. diameter of injector nozzle for injectable air, diameter of ring
US
nozzle for grinding air, and air pressure for milling). The fifth sample (SD) included respirable
AN
spray-dried amiloride HCl particles prepared from aqueous feed solutions under controlled
conditions (feed concentration, feed rate, atomization pressure, inlet temperature, outlet
M
temperature) to obtain size distribution comparable with jet-milled particles. The key properties
ED
of the obtained samples relevant for CFD-DPM simulations of DPI performance (i.e., geometric
particle size expressed as d10, d50, d90, powder density) are depicted in Table 1. Geometric PSDs
PT
were determined by laser diffraction, and true density of the samples was measured using helium
CE
pycnometry. Djokic et al. (2014) also examined aerosol dispersion characteristics with eightstages ACI with a pre-separator using the Aerolizer® (Pfizer) DPI device filled with 20 mg drug.
AC
Methods of determination are described in detail in the published study (Djokic et al., 2014).
Relevant parameters calculated based on ACI measurements (i.e. emitted dose (ED), fine particle
fraction (FPF), MMAD and GSD values) are also depicted in Table 1. These data were used for
comparison with the values estimated based on the CFD-DPM results.
2.2. CFD simulations
10
ACCEPTED MANUSCRIPT
2.2.1 3D Geometric models
In order to simulate powder dispersion in a commercial inhaler, three-dimensional model of DPI
had to be made. Aerolizer® model was created using commercial software for computer aided
design based on the geometry of the Aerolizer® device (Fig. 1) which was constructed from
T
detailed measurements taken from the marketed device using a micrometer. Some simple model
IP
modifications were added in order to remove chamfered and curved edges. These edges were
CR
ignored in areas where that do not have any effect on the process of inhalation. From this model
we have taken the inner cylinder and inner compartment where air and drug particles flow (fluid
US
domain). This fluid domain model (Fig. 2) consists of a chamber with two inlets, a barrel, a grid
AN
in between and a capsule for powder storage. This model was exported as .stp file that was later
used for mesh creation and to define boundary conditions. Meshed model is shown in Fig. 3.
M
Unstructured (tetrahedral) mesh was used. The tetrahedral cells were then converted to
ED
polyhedral cells to reduce the skewness of the overall cell count. The model grid consists of 281
782 polyhedral cells and 1 402 608 nodes. The grid was refined in the areas with expected large
CE
PT
gradients of flow and then additionally refined near the wall based on the initial solutions.
2.2.2 Numerical Simulation
AC
In order to precisely describe what happens during inhalation in the device and in large airways
we need to use turbulent flow. The commercial CFD software ANSYS Fluent 16.0 was used to
simulate fluid flow which is governed by the Navier–Stokes equations:
(
 2 



 2 
+ 
+ (
+
)
)=−



   
Eq. (1)
11
ACCEPTED MANUSCRIPT
The boundary conditions of air inlets, outlet of mouthpiece and other walls in the computational
models were set as velocity inlet, outflow and wall, respectively. To simplify conditions in the
model, it was assumed that:
(1) air in the model was incompressible;
T
(2) flow in the inlets was steady and normal to the surface area of the inlet;
IP
(3) the friction heat was negligible in this process.
CR
According to Milenkovic et al. (2013), during forced inhalation the instantaneous volumetric
flow rate rapidly increases and reaches its highest value. This instantaneous flow rate is
US
approximately equal to the highest value during inhalation and this leads to the possibility to
AN
approximate dynamic airflow as steady airflow which was done here.
In general, the k-ε model or the k-ω model can be used for inhalation problems. Launder and
M
Spalding (1974) as well as Menter (1994) have provided more details on these turbulence
ED
models. Many researchers have used the k-ω or Menter’s Shear Stress Transport (SST) models.
Versteeg et al. (2000), Longest et al. (2007), Kleinstreuer et al. (2007), Longest et al. (2008),
PT
Longest et al. (2009) have used this model because of its capability to be applied throughout the
CE
turbulent boundary layer. Therefore, Standard k-ω model with low Reynolds number (LRN) was
adopted for the turbulent flow of air as fluid. The k-ω turbulent model is a two-transport-
AC
equation model for turbulence prediction based on two partial difference equation for two
variables (kinetic energy k and specific turbulent dissipation rate or turbulent frequency ω). This
model has shown good accuracy when combined with high efficiency in comparison to complex
methods, such as large eddy simulation (LES) (Longest et al., 2013). The LRN k-ω model can be
used for accurate prediction of pressure drop, velocity profiles, and shear stress for transitional
and turbulent flows (Ghalichi et al., 1998).
12
ACCEPTED MANUSCRIPT
All transport equations were discretized to be at least second order accurate in space. A SecondOrder Upwind scheme was used to discretize the pressure equations, and a Quadratic Upwind
Interpolation for Convection Kinetics scheme was used to discretize the momentum and
turbulence equations. Both of these schemes provide second-order accuracy (Shur et al., 2012).
T
The numerical equations were converged to a steady-state solution. Rosin-rammler diameter
IP
distribution of inert particles, injected using face normal direction, was used as input for Fluent's
CR
Discrete Phase Model (DPM). Collisions of particles with the DPI walls occur predominantly
due to inertial impaction. The particle-wall collision frequency and capture efficiency, which
US
determine the rate of deposition, are controlled by adhesion forces (Milenkovic et al., 2014a).
AN
We simulated the case when powder instantaneously breaks-up into a population of particles and
aggregates identical to that of the free-flowing powder after which no more breakage or
M
aggregation occurs. Although this approach represents a limiting case of weak cohesion forces, it
ED
provides a means for evaluating the effects of flow rate, particle size, and adhesion forces on the
local and total particle deposition in the DPI device. With DPM, each particle is presented by a
PT
parcel of particles and this method neglects inter-particle collisions (Kloss et al., 2009). Due to
CE
the number of different simulations performed, it was not feasible to use additional tools such as
Discrete Element Method (DEM), which considers collision breakage or particle – particle
work.
AC
collision. This approach needs greater CPU effort (Kloss et al., 2009)and will be left for future
It was assumed that particles behave in such way that in contact with wall they stay „trapped“
and the particles that reach the outlet surface „escape“ the device domain. This assumption is
valid due to the fact that normal component of the particle velocity at the moment of impact with
13
ACCEPTED MANUSCRIPT
the inhaler wall ( ) is smaller than the critical normal velocity throughout the simulation ( ).
Detailed impaction model calculation can be found in Milenkovic et al., 2013.
In that way, it could be calculated how many particles were set at the beginning of the
calculation and how many particles got out of the device.
T
In order to solve the particle trajectories the spherical nonlinear drag law implemented in
IP
ANSYS Fluent (version 6.3, ANSYS, USA) was used. For a low Reynolds number, mentioned
CR
law reduces to Stokesian drag law. The gravitational acceleration was not included throughout
US
the simulations. It was shown that the gravity has the insignificant impact due to small aerosol
diameter sizes and short residence times in the inhaler itself (Milenkovic et al., 2013). It is
AN
assumed that all particles were released at the same time (beginning of simulation) using RosinRammler distribution on the inlet surface. Particle size is considered to be constant throughout
M
simulation. In general, particle cohesion forces affect the initial powder release and dispersion
ED
dynamics as well as aggregate deposition and breakage, as observed and explained in Milenkovic
et al. (2014b). Interactions between particle or aggregate with the inhaler walls, other particles or
PT
aggregates is very important. Outcome of these interactions is breakage of aggregates into
CE
smaller aggregates and fine particles which are important as they are best suited to target the
upper respiratory tract (Milenkovic et al. 2013, Milenkovic et al. 2014b). Boundary conditions
AC
for this simulation included velocity at the two inlets and pressure outlet (underpressure to
simulate suction effect which comes from the inhalation). The normalized Reynolds stress
residuals in the range of 10e−4 were applied as the convergence criteria to ensure full
convergence. Inlet velocity was calculated from the flow rate (28.3 l/min) and it was 11.79 m/s.
This flow rate was used because the aerodynamic PSD was experimentally measured at this
airflow rate (Djokic et al., 2014). Particle mass at the beginning was 20 mg. In total 5
14
ACCEPTED MANUSCRIPT
simulations were performed, 4 for jet-milled particles (JM1, JM2, JM3 and JM4) and 1 for spraydried particles (SD). Diameter and density values used for these simulations are given in Table 1.
2.3. Aerodynamic particle size distribution of CFD-DPM generated aerosol particles
T
The results of CFD-DPM numerical simulations for the five model formulations, namely
IP
geometric diameter, density and mass of the particles leaving the inhaler, were used to quantify
CR
aerodynamic PSDs of the generated aerosol by calculating the corresponding MMAD and GSD
values. Aerodynamic particle diameter (daerodynamic) was calculated from the following
US
expression:
Eq. (2)
AN
d aerodynamic  d geometric  
M
where dgeometric and ρ are geometric particle diameter and particle density, respectively. MMAD
for each sample (JM1, JM2, JM3, JM4 and SD) was estimated from the plot representing
ED
percentage cumulative mass vs. cut-off aerodynamic diameter on a log-scale. The applied
PT
method was based on linear interpolation between the nearest data points on either side of the
cumulative 50th mass percentile value (Christopher et al., 2010). GSD was derived from the
d 84
d16
Eq. (3)
AC
GSD 
CE
estimated d16 and d84 for each sample using the equation (3):
where d16 and d84 are aerodynamic particle diameters at the 16th and 84th cumulative mass
percentiles. The estimated MMAD and GSD values based on CFD-DPM data, along with the
CFD-DPM predicted ED, were further used as inputs in the designed amiloride-specific PBPK
model, to explore model estimates on drug deposition and absorption in the lungs following
inhalation of the five model formulations.
15
ACCEPTED MANUSCRIPT
2.4. Drug-specific PBPK model
Amiloride absorption and concomitant pharmacokinetic processes following oral and inhalation
administration were simulated using GastroPlusTM software (version 9.0.0007, SimulationPlus
T
Inc., USA). The in silico study consisted of two phases: first model construction, and afterwards
IP
model exploration. The first phase referred to the selection of input parameters and evaluation of
CR
the simulation outcomes by comparison with the data from the in vivo studies. In the exploration
phase, the generated drug-specific PBPK model was used to forecast amiloride deposition and
US
absorption following inhalation of the model formulations (JM1-JM4, SD).
AN
Drug absorption was predicted using Advanced Compartmental Absorption and Transit (ACAT)
fasted-state model, linked with PBPK model to simulate drug disposition through different body
M
tissues, and PCAT™ model to account for drug bioperformance in the respiratory tract. The
ED
ACAT model of human gastrointestinal (GI) tract is comprised of nine compartments linked in
series (stomach, duodenum, two jejunum compartments, three ileum compartments, caecum, and
PT
ascending colon), which are further subdivided to account for the drug that entered into the
CE
enterocytes. The effect of physiological conditions on drug absorption as it transits through
successive GI compartments is managed by a series of differential equations (Agoram et al.,
AC
2001). The drug was assumed to substantially bind in enterocytes (95%), to account for the
delayed time to reach maximum plasma concentration (tmax) (Savic et al., 2007). This
phenomenon was observed in case of other basic drugs with high solubility/fast dissolution and
high permeability (Sarfraz et al., 2015; Samanth et al., 2017).
In extension to the ACAT model, PBPK model enabled simulation of drug distribution in major
tissues linked together by blood circulation (Theil et al., 2003; Kostewicz et al., 2014). Each
16
ACCEPTED MANUSCRIPT
tissue was described by a set of physiological parameters, whereas default parameters for the
selected "30 years/70 kg" physiology were generated by the software, based on population mean
values obtained from published data. All of the tissues were considered to be well-stirred, and
drug distribution was treated as perfusion limited. Tissue to plasma partition coefficients (Kp
T
values) were calculated using the default Rodgers-Single (Lukacova) Kp method (Lukacova et
IP
al., 2008), and further optimized (multiplied by 3.5 scaling factor) to capture the experimental
CR
data on drug volume of distribution (Vd). Amiloride is predominantly excreted by the kidneys
via tubular secretion, which signifies relatively high renal clearance values (approximately 3
US
times the creatinine clearance) (Spahn et al., 1997). The estimated value for model development
AN
of 25 l/h was in the range of values reported in literature (Spahn et al., 1997; Fagerholm, 2007;
Savic et al., 2007).
M
In case of drug dosing via respiratory route, PCAT™ model was used to describe drug transit
ED
and absorption through the following pulmonary compartments: extra-thoracic (nasopharynx,
oropharynx, larynx), thoracic (trachea, bronchi), bronchiolar (bronchioles, terminal bronchioles),
PT
and alveolar-interstitial (respiratory bronchioles, alveolar ducts and sacs, interstitial connective
CE
tissue) (Chaudhuri and Lukacova, 2010). Deposition of inhaled particles in each compartment
was calculated by the software built-in ICRP 66 deposition model (ICRP, 1994). Nasal-
AC
pulmonary physiology parameters were kept at software default values. The additional input
parameters required for the simulation were obtained from literature or in silico estimated (Table
2).
In the model construction phase, published data from amiloride pharmacokinetic studies were
used to evaluate the generated drug PBPK model. Digital extraction of literature data from the
graphs was performed using DigItTM program (version 1.0.4, Simulations Plus, Inc., USA).
17
ACCEPTED MANUSCRIPT
Amiloride-specific PBPK model for oral drug delivery was validated by comparing the
simulation results with drug plasma concentration-time profiles following administration of oral
solution (Jones et al., 1997), or single amiloride HCl immediate (IR)-release capsule (Weiss et
al., 1969). The predictive power of the adjusted pulmonary drug delivery model was assessed by
T
comparing the simulation outcomes with the in vivo data on amiloride plasma profiles following
IP
pulmonary delivery of the nebulized solution (Jones et al., 1997). In order to simulate the
CR
deposition pattern of the nebulized solution, we used in vitro obtained MMAD value from
literature as input. We need to note that the proper way to validate the applied CFD-PBPK
US
modelling approach would be to use CFD simulated MMAD and GSD as input values. But, due
AN
to the lack of relevant CFD data for this nebulized solution, an in vitro obtained MMAD was
used. Another annotation regarding the construction of the PCAT™ model is that drug systemic
M
absorption rate constants in the pulmonary compartments (Ka(p)) were varied to assess the
ED
influence of the absorption kinetics in the lungs on the generated pharmacokinetic profiles.
Predictability of the generated models (oral and inhalation) was measured by the percent
observed  predicted
 100
observed
Eq. (4)
CE
% PE 
PT
prediction error (%PE) between the simulated and in vivo observed data:
AC
In the model exploration phase, the constructed model was used to estimate the expected drug
deposition and absorption following inhalation of five model formulations (JM1- JM4, SD). In
this step, MMAD and GSD values calculated based on the aerodynamic aerosol particle
distribution on the outlet of the Aerolizer® model, obtained by CFD-DPM, were used as inputs
in the PBPK model. Also, CFD-DPM predicted ED for each model formulations was used to
define drug dose, and drug dissolution in the lungs was predicted based on CFD predicted
geometric PSD and input drug solubility.
18
ACCEPTED MANUSCRIPT
3. Results and Discussion
3.1. CFD simulation results
Results for jet-milled samples and spray-dried sample are similar with small differences in peak
T
values. In order not to repeat figures, we will only show results for JM2 sample.
IP
Fig. 4 shows the velocity magnitudes, tangential and radial velocities for xz plane obtained from
CR
simulation using previously mentioned parameters. Flow pattern in the inhaler is complex, due to
swirling motion inside the capsule chamber that incoming flow generates. The highest velocity is
US
observed inside the grid, i.e., about 48 m/s, while the lowest one is in the barrel. Obtained
AN
velocity values and distribution can be compared to values in other published papers (Tong et al.,
2013).
M
Fig. 5 shows the velocity distribution in the grid area. It can be noticed that grid suppress the
ED
turbulence in the barrel and also reduces swirl that is generated in the capsule chamber. Higher
air inlet velocity leads to wider impact velocity and higher turbulence levels. Presented results
PT
are similar to the results obtained by Coates et al. (2006).
CE
Maximum velocity values are in range from 45.23 m/s (JM3 and JM4) to 48.81 m/s (JM2 and
SD). Fig. 6 shows the outlet velocity vectors for jet-milled (JM1, JM2, JM3, JM4) and spray-
AC
dried particles (SD), which is helpful in determination of particles deposition in the oral cavity.
As seen in Fig. 7, the fluid flow inside the Aerolizer evokes particle collisions with the inhaler
wall, especially in the area of carrier and chamber before grid structure. It should be noted that
particle–particle collisions generally have to be taken into account. However, when the particle
volume fraction is less than about 10-3 (Sommerfeld et al., 2008, Milenkovic et al., 2014) this can
be ignored. Therefore, due to the small volume fraction (10-4) in this work, the effect of particle –
19
ACCEPTED MANUSCRIPT
particle collisions were not considered or taken into account. Most of the carrier-device
collisions as well as carrier-carrier collisions occur in the chamber. Carrier-device collisions are
more common than carrier-carrier collisions. This type of collision (carrier-carrier) rarely
happens in the barrel because the flow in the barrel is less turbulent than the flow in the chamber
T
and also because the particles enter the barrel at the different times and reduce the chance for
IP
collision. However, the main reason for why carrier-carrier collisions rarely occur is the very
CR
small particle volume fraction. The particle-particle collision rate depends on the second power
of the local particle number density function. Studies such as Li and Ahmadi (2007), Ge et al.
US
(2012) and Zhou et al. (2013) have used similar conditions regarding particle collision where
AN
rebound was ignored. Obtained results can be compared to the result obtained by Tong et al.
(2015). This swirling flow reduces and becomes less dominant through the mouthpiece, as the
M
tangential air inlets which introduce a swirl component, don’t have such great impact down the
PT
3.2. Particle tracking results
ED
flow after the grid structure.
CE
The particle tracking was initially carried out to simulate the dispersion of 1000 particles and
repeated for 5000, and 10000 particles to determine if the particle impaction velocities obtained
AC
were independent of the number of particles simulated. Since the velocities proved to be
independent of the number of particles, we could continue working with smaller number of
particles, since the simulation time and necessary computer memory drastically increases with
particle number.
Particles mostly get “trapped” inside the carrier area because of the swirl flow. This happens
after a very short period, showing that the residency time in this area is short. Fig. 8. shows the
20
ACCEPTED MANUSCRIPT
results of particle tracking from all jet-milled particles as well as spray dried particles. In order to
have better visualization only 20 tracks were shown. Particles at the exit of the barrel are
concentrated in the middle.
T
3.3. Comparison of in vitro data and in silico predictions for the model DPI formulations
IP
Comparative analysis of the in vitro obtained data (Djokic et al., 2014) and in silico CFD-DPM
CR
predictions revealed similar values for the percent ED (Fig. 9a). In the case of JM1, almost
identical values were obtained in vitro and in silico. An excellent match was also seen for the
US
other jet-milled samples (difference in ED up to 4%), whereas for spray-dried sample the
AN
obtained difference between these values was about 4.5%. Such results demonstrate that the
generated DPI inhaler model adequately describes aerosolization of five model formulations, and
M
can be used to calculate ED of an aerosolized DPI.
ED
According to the simulation outcomes, the ED from all tested formulations is comprised of
particles with the aerodynamic diameter of less than 5 microns, and therefore can be considered
PT
as FPF. In contrast, the values resulting from ACI measurements indicate that only about 40 to
CE
50% of the emitted particles mass correspond to FPF, depending on the formulation (Table 1).
The differences in FPF values based on CFD-DPM simulations and ACI measurements may
AC
arise from the presence of agglomerates which are detected by the ACI device, in addition to the
fine particle adhesion (and loss) onto inner ACI walls, while the applied in silico modelling
follow the trajectories of single particles. This implies that both methods have certain limitations.
The generated CFD-DPM model does not take into account particle agglomeration that most
likely happens prior to and during aerosolization, and therefore the estimated FPF values might
be overestimated. On the other hand, during ACI measurements drug particles may also
21
ACCEPTED MANUSCRIPT
agglomerate and de-agglomerate during fractionation (Strickland et al., 2013), meaning that the
obtained results do not necessarily reflect just the influence of agglomerates at the inhaler outlet.
Primary PSDs resulting from CFD-DPM simulations are depicted in Fig. 10. The presented data
demonstrate differences between the predicted size distribution of particles leaving the inhaler
T
for different test formulations. Cumulative distribution data (Fig. 10) were used to calculate the
IP
corresponding aerodynamic PSDs, and the derived MMAD and GSD values were compared with
CR
the values calculated from DPI PSD data measured using ACI (Fig. 9b,c). In general, the
simulations indicated much lower MMAD in comparison to the values calculated based on ACI
US
measurements. The largest divergence between the in vitro and in silico-based MMAD values
AN
was observed for JM1 and JM2 samples, but notable difference were seen for other samples, too.
As discussed by Djokic et al. (2014), relatively large MMAD for JM1 and JM2 sample,
M
calculated based on ACI analysis, may result from particle tendency to form aggregates. In
ED
addition, adhesion of particles to the impactor walls might also influence the obtained in vitro
results (de Boer et al., 2002). Similar observations were made for GSD where CFD-DPM
PT
generated values indicated more narrow aerosol size distribution, specially in the case of JM1
CE
and JM2 samples. These findings imply that the aerosol cloud that leaves the inhaler might
contain smaller particles, with narrower size distribution, than assumed based on ACI
AC
measurements. Therefore, CFD-DPM generated results, in conjunction with the in silico model
that describes particles transport through the lungs (e.g. ICRP66 model (ICRP, 1994)), can be
used as an alternative to the in vitro ACI assessment of DPI deposition pattern in vivo. Still, we
need to note that the described CFD-DPM approach may lead to an underestimation of MMAD
when a significant portion of particles leaves the inhaler as agglomerates, which is often the case
for DPIs. We should also note that the assumption of a 100% sticking efficiency of large
22
ACCEPTED MANUSCRIPT
particles or aggregates could contribute to the underestimation of MMAD as this would
effectively eliminate the large particle sizes from the in silico predicted PSD. In order to
accurately describe properties of the inhaled aerosols, simulation tools need to address
complicated particle-particle and particle-wall interactions, including agglomeration and de-
T
agglomeration.
IP
Fig. 11 shows CFD-DPM predicted percentage of deposited particles in different deposition sites
CR
(capsule chamber, circulation chamber, grid area and mouthpiece) for the corresponding Q=28.3
l/min. The presented simulation results correspond to each of the powder types JM1-JM4 and
US
SD. It can be observed that most particles deposit in either capsule or circulation chamber, whilst
AN
only a small percentage of particles deposit in the mouthpiece.
M
3.4. PBPK model construction
ED
After inhalation, amiloride enters the systemic circulation in two phases: quickly via respiratory
tract, and more slowly following GI absorption of the swallowed part of the dose (Jones et al.,
PT
1997; Nooney et al., 1997). Furthermore, in patients with cystic fibrosis, it could be expected that
CE
a larger portion of aerosolized particles will deposit in the upper lung regions (due to decreased
lung capacity in comparison to healthy subjects), and this portion will eventually be swallowed.
AC
Therefore, drug-specific PBPK model was first developed and validated for oral drug dosing,
and further adjusted to simulate plasma profile following inhalation. A simpler compartmental
model might also have been used for amiloride pharmacokinetic modeling, but a whole body
PBPK was considered advantageous because of its wider applicability. Namely, it can be used in
future studies to forecast drug-drug or drug-disease (in terms of physiological conditions)
interactions.
23
ACCEPTED MANUSCRIPT
The simulation results for 10 mg amiloride oral solution and 20 mg IR oral capsule are shown in
Fig. 12 and Table 3. The two profiles were based on the same input data (Table 2), with the only
differences concerning the selection of drug dose, dosage form, and concomitant simulation of
drug dissolution from a capsule. Due to high solubility of amiloride HCl, the predicted drug
T
dissolution from IR capsule happened almost immediately (100% drug dissolved in 5 min),
IP
which resulted in the same tmax for both formulations.
CR
The predicted pharmacokinetic parameters for 10 mg amiloride oral dose fitted into the range of
values reported in literature (Jones et al., 1997; Spahn et al., 1987). The simulation results for 20
US
mg drug dose also agreed with the in vivo data, as demonstrated by %PE values less than 10%
AN
for tmax and AUC (Table 3). In this case, larger deviation from the mean in vivo data (Weiss et
al., 1969) was noted solely for Cmax, but considering the range of values observed in different
M
studies (Weiss et al., 1969; Grayson et al., 1971), the simulated value can be considered as a
ED
reasonable estimate. In addition, the estimated Vd value of 577.97 l agreed well with the large
volume of distribution reported in literature (Grayson et al., 1971; Savic et al., 2007), and the
PT
predicted elimination half-life (t1/2 = 16.02 h) matched the in vivo observed values (Jones et al.,
CE
1997; Spahn et al., 1987), indicating that the selected PBPK model parameters adequately
describe amiloride pharmacokinetics.
AC
In the next step, the generated model was adjusted to predict the absorption of a nebulized
amiloride solution, at 1.91 mg dose level. This was the estimated dose delivered to the lungs,
determined from the initial dose (4.5 mg) minus the amount of drug entrapped in the nebulizer
(Jones et al., 1997). The additional adjustments in the model parameters included dosage form
(pulmonary solution), and mean particle size (in vitro obtained MMAD = 4 μm taken from
literature (Jones et al., 1997)). In addition, it was assumed that fraction of the inhaled drug (from
24
ACCEPTED MANUSCRIPT
extra-thoracic region) would be swallowed, and in order to match the in vivo data (Jones et al.,
1997) gastric residence time was prolonged to 1 h to simulate the scenario where it takes some
time for the swallowed portion of the drug to reach gastrointestinal tract.
The obtained results are presented in Fig. 13 and Table 3. Initial simulation (model M1) provided
T
plausible prediction of the inhaled amiloride absorption. However, this model was based on the
IP
default software calculated Ka(p) for amiloride of 8.42 x 10-3 1/s (under default settings, a uniform
CR
value is used for all pulmonary compartments), which may not be considered adequate in the
case of a basic compound like this. Namely, default Ka(p) is estimated from the lung volume,
US
blood flow, drug tissue to plasma partition coefficient and blood to plasma concentration ratio
AN
(GastroPlusTM manual, 2015), and it does not take into account specific phenomena that may
happen in vivo. Some recent studies, using a model based population pharmacokinetic approach
M
or in vitro lung slice methodology, have demonstrated that certain basic drugs (e.g. olodaterol,
ED
some β2-sympathomimetics) exhibit prolonged pulmonary residence times due to lysosomal
trapping in the lungs and thus delayed absorption (Backstrom et al., 2016; Borghart et al., 2016).
PT
It was further noted in these reports that drug absorption from the lungs can be described as
CE
parallel absorption processes with different rate constants, whereas the initial fast absorption is
responsible for the early Cmax, while slow absorption (a major fraction) becomes apparent in the
AC
terminal phase and may contribute to the prolonged therapeutic effect. Furthermore, these
different absorption processes can take place in the same lung region, e.g. fast and slow
absorption from the alveolar space as demonstrated for the inhaled olodaterol (Borghart et al.,
2016), meaning that they could not be matched with distinct pulmonary compartments.
In order to assess the influence of decreased lung absorption rate on the generated
pharmacokinetic profiles for amiloride, we have reduced the initial Ka(p) values by 25% (model
25
ACCEPTED MANUSCRIPT
M2), 50% (model M3) and 75% (model M4). According to the obtained results, the reduction in
Ka(p) had no influence on the extent of amiloride absorption (the same predicted AUC values),
but there were noticeable differences in the predicted Cmax, and especially tmax values (Table 3).
The best match between the in vivo and predicted data was observed with model M3, where the
T
initial Ka(p) was decreased by 50%. These results indicate that inhaled amiloride, similar to other
IP
basic drugs, exhibit certain delay in pulmonary absorption. Increasing the extent of Ka(p)
CR
reduction leads to the loss of the early peak in plasma concentration-time profile, as
demonstrated with the model M4 (Fig. 13). Therefore, model M3 was adopted for further
US
exploration. It could still be speculated whether this model best describes inhaled amiloride
AN
pharmacokinetics because of the inability to interpret possible parallel (fast and slow) pulmonary
absorption processes. Namely, in the current PCAT™ model only one Ka(p) value can be ascribed
M
to each pulmonary compartment. In this context, upgrades in the software PCAT™ model might
ED
improve in silico predictions for this drug.
According to the data in Table 3, the predictive ability of M3 model seems rather good. %PE for
PT
AUC, calculated based on mean in vivo data, was somewhat high; however, the predicted value
CE
fitted into the range of individually recorded values (Jones et al., 1997). Moreover, certain
deviations from the mean in vivo observed profile might be ascribed to the lack of adequate input
AC
data regarding PSD. Namely, the authors (Jones et al., 1997) only specified MMAD, and stated
that the target site for this heterodispersed aerosol were the conducting airways. The predicted
regional absorption distribution indicated that about half of the systemically available amiloride
following inhalation of the nebulized solution is absorbed through the lungs (15.8%), while the
rest of the dose is absorbed from the gastrointestinal tract (Fig. 14). These results are in
accordance with the findings of Jones et al. (1997), demonstrating that the generated model M3
26
ACCEPTED MANUSCRIPT
was able to interpret the absorption pattern of nebulized amiloride, including the second peak
phenomenon.
3.5. PBPK model exploration
T
The constructed amiloride-specific absorption model (model M3) was consequently used to
IP
predict pulmonary deposition, and systemic drug absorption from different amiloride DPI
CR
experimental formulations. As already stated, these formulations had different aerodynamic
properties, and relevant attributes (i.e., CFD-DPM predicted delivered dose and particle size)
US
were accounted for in the input parameters datasets. Aerodynamic PSD for each formulation,
AN
estimated based on CFD-DPM results, was used to forecast drug deposition through different
pulmonary compartments. The simulated deposition profiles are depicted in Fig. 15. It can be
M
noted that drug deposition fractions in peripheral region of the lungs will be similar for all
ED
formulations. More pronounced differences are seen in drug deposition in the conducting
airways, whereas formulations with higher MMAD values (JM3 and JM4) will have higher
PT
percentage of drug deposited in these regions. In contrast, in the case of very small particles with
CE
MMADs ≤ 2 μm (formulations JM1, JM2 and SD), more than 50% of the delivered dose will be
exhaled, under normal physiological conditions and breathing pattern as specified in the model.
AC
The corresponding plasma concentration-time profiles for the five formulations are shown in Fig.
16, along with the data on pulmonary drug absorption. Since majority of inhaled amiloride is
absorbed through the lung compartments (smaller portion is swallowed and absorbed from the
gastrointestinal tract), extent of systemic drug bioavailability is a reflection of pulmonary
absorption. In the case of JM1 and JM2, drug absorption is expected to be additionally decreased
because aerosol delivery (in terms of delivered dose) is less efficient, as outlined in Fig. 9a.
27
ACCEPTED MANUSCRIPT
As already mentioned, one of the benefits associated with PBPK modelling is prediction of drug
absorption under different physiological conditions. Since cystic fibrosis is usually associated
with impaired airflow (Zapletal et al., 1993), sensitivity analysis was run to investigate the effect
of this parameter on the expected amiloride regional lung deposition. Simulated data for the
T
formulation JM4, depicted in Fig. 17, illustrate how decreased airflow can decrease amiloride
IP
deposition in the conducting airways, which are the target site for this drug action. Such results
CR
suggest that drug & patient-specific PBPK modelling can provide quantitative data on the
expected drug concentration in target site of the lungs, and thus enable the selection of
AN
US
appropriate dosing regimen.
4. Conclusion
M
The particle dispersion in a DPI was simulated by a CFD-DPM model. Boundary conditions that
ED
included velocity inlet and pressure outlet were set based on experiments of Djokic at al. (2014).
Input data on particle characteristics were taken from the same study. For the corresponding
PT
turbulence model, k-ω model was used, and it was assumed that particles behave in such a way
CE
that in contact with wall they stay “trapped”, and the particles that reach the outlet surface
“escape” the device domain. Under this assumption, it was possible to calculate how many
AC
particles were set at the beginning of the calculation, and how many particles got out of the
device. Analysis of the flow field using CFD analysis provided information about air velocity
during inhalation. Based on this flow field, we were able to determine how drug particles are
moving from the capsule chamber until the end of the barrel, and particle distribution present on
the outlet. The results indicate that particles mostly get „trapped“ inside the carrier area because
of the swirl flow, and this swirling flow becomes less dominant through the mouthpiece, as the
28
ACCEPTED MANUSCRIPT
tangential air inlets which introduce a swirl component, do not have such great impact down the
flow after the grid structure. CFD-DPM generated values for the percent ED are within 4.5%
difference compared to the experimental data obtained by Djokic at al. (2014). Results presented
in the CFD-DPM part here are obtained with simplification that included only particles that were
T
placed at the bottom of the chamber area and did not include the capsule itself. Also, the
IP
constructed model did not account for particle interactions such as agglomeration and de-
CR
agglomeration, and these phenomena will be included in the upgraded version of the model.
Future work will include new simulations with inhaler where particles will be inside of a capsule,
US
and the results will be used as input in the analysis that includes particle tracking inside patient’s
AN
lungs.
Although computationally-intensive, this approach can be seen as useful alternative to the in
M
vitro determination of aerosolized particle properties using ACI or other types of impactors.
ED
Linked with PBPK modeling, these data enable prediction of drug absorption in specific patients,
under different physiological conditions, which may facilitate the decision on efficient dose
PT
levels for specific patients. The combined CFD-PBPK modeling might be seen as a promising
CE
tool in the design and characterization of DPIs; however, this approach needs to be additionally
AC
tested with different DPI types, and different model drugs.
Acknowledgements
This study is based upon work from COST Action MP1404 SimInhalle 'Simulation and
pharmaceutical technologies for advanced patient-tailored inhaled medicines', supported by
COST (European Cooperation in Science and Technology) www.cost.eu. In addition, this work
29
ACCEPTED MANUSCRIPT
was supported by the Ministry of Education, Science and Technological Development, Republic
of Serbia with projects TR34007, III41007 and OI174028.
References
T
Agoram, B., Woltosz, W., Bolger, M., 2001. Predicting the impact of physiological and
IP
biochemical processes on oral drug bioavailability. Adv. Drug Deliv. Rev. 50, S41–S67.
CR
Almukainzi, M., Jamali, F., Aghazadeh-Habashi, A., Löbenberg, R., 2016. Disease specific
modeling: Simulation of the pharmacokinetics of meloxicam and ibuprofen in disease state
US
vs. healthy conditions. Eur. J. Pharm. Biopharm. 100, 77–84.
AN
Alqahtani, S., Kaddoumi, A., 2016. Development of a Physiologically Based
Pharmacokinetic/Pharmacodynamic Model to Predict the Impact of Genetic Polymorphisms
M
on the Pharmacokinetics and Pharmacodynamics Represented by Receptor/Transporter
ED
Occupancy of Central Nervous System Drugs. Clin. Pharmacokinet. 55(8), 957–969.
Backman, P., Tehler, U., Olsson, B., 2016. Predicting exposure after oral inhalation of the
PT
selective glucocorticoid receptor modulator, AZD5423, based on dose, deposition pattern,
CE
and mechanistic modelling of pulmonary disposition. J. Aerosol. Med. Pulm. Drug Deliv.
30(2), 108–117.
AC
Backstrom, E., Boger, E., Lundqvist, A., Hammarlund-Udenaes, M., Friden, M., 2016. Lung
retention by lysosomal trapping of inhaled drugs can be predicted in vitro with lung slices.
J. Pharm. Sci. 105(11), 3432–3439.
Borghardt, J.M., Weber, B., Staab, A., Kloft, C., 2015. Pharmacometric models for
characterizing the pharmacokinetics of orally inhaled drugs. AAPS J. 17(4), 853–70.
30
ACCEPTED MANUSCRIPT
Borghardt, J.M., Weber, B., Staab, A., Kunz, C., Formella, S., Kloft, C., 2016. Investigating
pulmonary and systemic pharmacokinetics of inhaled olodaterol in healthy volunteers using
a population pharmacokinetic approach. Br. J. Clin. Pharmacol. 81(3), 538–552.
Bouzom, F., Ball, K., Perdaems, N., Walther, B., 2012. Physiologically based pharmacokinetic
T
(PBPK) modelling tools: how to fit with our needs? Biopharm. Drug Dispos. 33(2), 55–71.
IP
Brocklebank, D., Ram, F., Wright, J., Barry, P., Cates, C., Davies, L., Douglas, G., Muers, M.,
CR
Smith, D., White, J., 2001. Comparison of effectiveness of inhaler devices in asthma and
chronic obstructive airway disease: a systematic review of the literature. Health Technol.
US
Assess. 5(26), 1–149.
AN
Chan, H.K., 2006. Dry powder aerosol delivery systems: current and future research directions.
J. Aerosol. Med. 19(1), 21–27.
M
Chaudhuri, S.R., Lukacova, V., 2010. Simulating delivery of pulmonary (and intranasal)
ED
aerosolised drugs. Orally Inhaled Nasal Drug Prod, 26-30.
Chetty, M., Rose, R.H., Abduljalil, K., Patel, N., Lu, G., Cain, T., Jamei, M., Rostami-Hodjegan,
PT
A., 2014. Applications of linking PBPK and PD models to predict the impact of genotypic
CE
variability, formulation differences, differences in target binding capacity and target site
drug concentrations on drug responses and variability. Front. Pharmacol. 5, 258.
AC
Christopher, J.D., Dey, M., Lyapustina, S., Mitchell, J.P., Tougas, T.P., Van Oort, M.,
Strickland, H., Wyka, B., 2010. Generalized simplified approaches for mass median
aerodynamic determination. Pharmacopeial Forum. 36(3), 812–823.
Clark, A.R., 2004. Pulmonary delivery technology: recent advances and potential for the new
millennium. A.J. Hickey (Ed.), Drugs and the Pharmaceutical Sciences, Marcel Dekker,
Inc., New York. 571–591.
31
ACCEPTED MANUSCRIPT
Coates, M.S., Fletcher, D.F., Chan, H.K., Raper, J.A., 2004. Effect of design on the performance
of a dry powder inhaler using computational fluid dynamics. Part 1: grid structure and
mouthpiece length. J. Pharm. Sci. 93(11), 2863–2876.
Coates, M.S., Chan, H-K., Fletcher, D.F., Raper, J.A., 2005a. Influence of air flow on the
T
performance of a dry powder inhaler using computational and experimental analyses.
IP
Pharm. Res. 22(9), 1445–1453.
CR
Coates, M.S., Fletcher, D.F., Chan, H-K., Raper, J.A., 2005b. The role of capsule on the
performance of a dry powder inhaler using computational and experimental analyses.
US
Pharm. Res. 22(6), 923–932.
AN
Coates, M.S., Chan, H-K., Fletcher, D.F., Raper, J.A., 2006. Effect of design on the performance
of a dry powder inhaler using computational fluid dynamics. Part 2: air inlet size. J. Pharm.
M
Sci. 95(6), 1382–1392.
ED
Coates, M.S., Chan, H.K., Fletcher, D.F., Chiou, H., 2007. Influence of mouthpiece geometry on
the aerosol delivery performance of a dry powder inhaler. Pharm. Res. 24(8), 1450–1456.
PT
Dahlgren, D., Roos, C., Sjögren, E., Lennernäs, H., 2015. Direct in vivo human intestinal
CE
permeability (Peff ) determined with different clinical perfusion and intubation methods. J.
Pharm. Sci. 104(9), 2702–2726.
AC
de Boer, A.H, Gjaltema, D., Hagedoorn, P., Frijlink, H.W., 2002. Characterization of inhalation
aerosols: a critical evaluation of cascade impactor analysis and laser diffraction technique.
Int. J. Pharm. 249(1-2), 219–231.
Djokic, M., Kachrimanis, K., Solomun, Lj., Djuris, J., Vasiljevic, D., Ibric, S., 2014. A study of
jet-milling and spray-drying process for the physicochemical and aerodynamic dispersion
properties of amiloride HCl. Powder Technol. 262, 170–176.
32
ACCEPTED MANUSCRIPT
Fagerholm, U., 2007. Prediction of human pharmacokinetics - renal metabolic and excretion
clearance. J. Pharm. Pharmacol. 59(11), 1463–1471.
Feng, Y., Kleinstreuer, C., 2014. Micron-particle transport, interactions and deposition in triple
lung-airway bifurcations using a novel modelling approach. J. Aerosol. Sci. 71, 1–15.
T
Fröhlich, E., Mercuri, A., Wu, S., Salar-Behzadi, S., 2016. Measurements of deposition, lung
IP
surface area and lung fluid for simulation of inhaled compounds. Front. Pharmacol. 7, 181.
CR
GastroPlusTM version 9.0 Manual, 2015. SimulationPlus Inc., Lancaster.
Ghalichi, F., Deng, X., Champlain, A.D., Douville, Y., King, M., Guidoin, R., 1998. Low
US
Reynolds number turbulence modelling of blood flow in arterial stenoses. Biorheology.
AN
35(4–5), 281–294.
Grayson, M.F., Smith, A.J., Smith, R.N., 1971. Absorption, distribution and elimination of 14 C-
M
amiloride in normal human subjects. Br. J. Pharmacol. 43(2), 473P–474P.
ED
Hoppentocht, M., Hagedoorn, P., Frijlink, H.W., de Boer, A.H., 2014. Technological and
practical challenges of dry powder inhalers and formulations. Adv Drug Deliv Rev. 75, 18–
PT
31.
CE
ICH, 2005. ICH Harmonised Tripartite Guideline Pharmaceutical Development Q8 (R2), Current
Step 4 version.
AC
https://www.ich.org/fileadmin/Public_Web_Site/ICH_Products/Guidelines/Quality/Q8_R1/
Step4/Q8_R2_Guideline.pdf (accessed 31.03.17).
ICRP, 1994. Human respiratory tract model for radiological protection. ICRP Publication 66.
Ann. ICRP 24 (1–3).
IPCS INCHEM. Amiloride Hydrochloride (Monograph 026).
http://www.inchem.org/documents/pims/pharm/pim026.htm (accessed 07.04.17).
33
ACCEPTED MANUSCRIPT
Islam, N., Gladki, E., 2008. Dry powder inhalers (DPIs) – a review of device reliability and
innovation. Int. J. Pharm. 360(1–2), 1–11.
Jiang, L., Tang, Y., Zhang, H., Lu, X., Chen, X., & Zhu, J. 2012. Importance of Powder
Residence Time for the Aerosol Delivery Performance of a Commercial Dry Powder
T
Inhaler Aerolizer®. J. Aerosol. Med. Pulm. Drug Deliv. 25(5), 265–279.
IP
Jones, K.M., Liao, E., Hohneker, K., Turpin, S., Henry, M.M., Selinger, K., Hsyu, P.H.,
CR
Boucher, R.C., Knowles, M.R., Dukes, G.E., 1997. Pharmacokinetics of amiloride after
inhalation and oral administration in adolescents and adults with cystic fibrosis.
US
Pharmacotherapy. 17(2), 263–270.
AN
Jones, H., Rowland-Yeo, K., 2013. Basic concepts in physiologically based pharmacokinetic
modeling in drug discovery and development. CPT Pharmacometrics Syst. Pharmacol. 2,
M
e63.
ED
Kleinstreuer, C., Shi, H., Zhang, Z., 2007. Computational analyses of a pressurized metered dose
inhaler and a new drug–aerosol targeting methodology. J. Aerosol Med. 20(3), 294–309.
PT
Kleinstreuer, C., Feng, Y., Childress, E., 2014. Drug-targeting methodologies with applications:
CE
A review. World J. Clin. Cases 2(12), 742–756.
Kloss, C., Goniva, C., Aichinger, G., Pirker, S., 2009. Comprehensive DEM-DPM-CFD
AC
simulations - model synthesis, experimental validation and scalability. Seventh International
Conference on CFD in the Minerals and Process Industries; CSIRO, Melbourne, Australia,
December 9–11.
Kostewicz, E.S., Aarons, L., Bergstrand, M., Bolger, M.B., Galetin, A., Hatley, O., Jamei, M.,
Lloyd, R., Pepin, X., Rostami-Hodjegan, A., Sjögren, E., Tannergren, C., Turner, D.B.,
34
ACCEPTED MANUSCRIPT
Wagner, C., Weitschies, W., Dressman, J., 2014. PBPK models for the prediction of in vivo
performance of oral dosage forms. Eur. J. Pharm. Sci. 57, 300–321.
Launder, B.E., Spalding, D., 1974. The numerical computation of turbulent flows. Comput.
Meth. Appl. Mech. Eng. 3(2), 269–289.
T
Longest, P.W., Hindle, M., Choudhuri, S.D., Byron, P.R., 2007. Numerical simulations of
IP
capillary aerosol generation: CFD model development and comparisons with experimental
CR
data. Aerosol Sci. Technol. 41(10), 952–973.
Longest, P.W., Hindle, M., Choudhuri, S. D., Xi, J., 2008. Comparison of ambient and spray
US
aerosol deposition in a standard induction port and more realistic mouth–throat geometry. J.
AN
Aerosol. Sci. 39(7), 572–591.
Longest, P.W., Hindle, M., Choudhuri, S.D., 2009. Effects of generation time on spray aerosol
M
transport and deposition in models of the mouth–throat geometry. J. Aerosol Med. Pulm. D
ED
22(2), 67–84.
Longest, P.W., Holbrook, L.T., 2012. In Silico Models of Aerosol Delivery to the Respiratory
PT
Tract – Development and Applications. Adv. Drug Deliv. Rev. 64(4), 296–311.
CE
Longest, P.W., Son, Y.J, Holbrook, L., Hindle, M., 2013. Aerodynamic Factors Responsible for
the Deaggregation of Carrier-Free Drug Powders to form Micrometer and Submicrometer
AC
Aerosols. Pharm. Res. 30(6), 1608–1627.
Lukacova, V., Parrott, N., Lave, T., Fraczkiewicz, M., Bolger, M., Woltosz, W., 2008. General
approach to calculation of tissue:plasma partition coefficients for physiologically based
pharmacokinetic (PBPK) modelling. AAPS National Meeting, Atlanta, November 15–20.
Menter, F.R., 1994. Two-equation eddy–viscosity turbulence models for engineering
applications. AIAA J. 32(8), 1598–1616.
35
ACCEPTED MANUSCRIPT
Milenkovic, J., Alexopoulos, A.H., Kiparissides, C., 2013. Flow and particle deposition in the
Turbuhaler: A CFD simulation. Int. J. Pharm. 448(1), 205–213.
Milenkovic, J., Alexopoulos, A.H., Kiparissides, C., 2014. Airflow and Particle Deposition in a
Dry Powder Inhaler: An Integrated CFD Approach. Simulation and Modeling
T
Methodologies, Technologies and Applications. Springer International Publishing, 127–
IP
140.
CR
Milenkovic, J., Alexopoulos, A.H. and Kiparissides, C., 2014. Deposition and fine particle
production during dynamic flow in a dry powder inhaler: A CFD approach. Int. J. Pharm.
US
461(1–2), 129–136.
AN
Mitchell, J.P., Nagel, M.W., 2003. Cascade impactors for the size characterization of aerosols
from medical inhalers: their uses and limitations. J. Aerosol. Med. 16(4), 341–377.
M
Mossaad, D.M.R., 2014. Drug Delivery to the Respiratory Tract Using Dry Powder Inhalers.
ED
Electronic Thesis and Dissertation Repository. Paper 2036.
Noone, P.G., Regnis, J.A., Liu, X., Brouwer, K.L., Robinson, M., Edwards, L., Knowles, M.R.,
PT
1997. Airway deposition and clearance and systemic pharmacokinetics of amiloride
1283–1290.
CE
following aerosolization with an ultrasonic nebulizer to normal airways. Chest. 112(5),
AC
Peters, S.A., 2008. Identification of intestinal loss of a drug through physiologically based
pharmacokinetic simulation of plasma concentration-time profiles. Clin. Pharmacokinet.
47(4), 245–259.
Samant, T.S., Lukacova, V., Schmidt, S., 2017. Development and qualification of
physiologically based pharmacokinetic models for drugs with atypical distribution behavior:
A desipramine case study. CPT Pharmacometrics Syst. Pharmacol. 6(5), 315–321.
36
ACCEPTED MANUSCRIPT
Sarfraz, M., Macwan, J., Almukainzi, M., Bolger, M., Loebenberg, R., 2015. The importance of
lysosomal trapping for setting clinically relevant product specifications for
dextromethorphan immediate release dosage forms. AAPS Annual Meeting and Exposition,
Orlando, October 25–29.
T
Savic, R.M., Jonker, D.M., Kerbusch, T., Karlsson, M.O., 2007. Implementation of a transit
CR
Pharmacokinet. Pharmacodyn. 34(5), 711–726.
IP
compartment model for describing drug absorption in pharmacokinetic studies. J.
Shur, J., Lee, S., Adams, W., Lionberger, R., Tibbatts, J., Price, R., 2012. Effect of Device
AN
Inhalers. The AAPS Journal 14(4), 667–676.
US
Design on the In Vitro Performance and Comparability for Capsule-Based Dry Powder
Sommerfeld, M., van Wachem, B., Oliemans, R., 2008. Best Practice Guidelines for
M
Computational Fluid Dynamics of Dispersed Multiphase Flows. ERCOFTAC European
ED
Research Community on Flow, Turbulence and Combustion
Spahn, H., Reuter, K., Mutschler, E., Gerok, W., Knauf, H., 1987. Pharmacokinetics of amiloride
PT
in renal and hepatic disease. Eur. J. Clin. Pharmacol. 33(5), 493–498.
CE
Strickland, H., Morgan, B., Mitchel, J.P., 2013. Physical Causes of APSD Changes in Aerosols
from OIPs and Their Impact on CI Measurements, in: Tougas, T.P., Mitchel, J.P.,
AC
Lyapustina, S.A. (Eds.), Good Cascade Impactor Practices, AIM and EDA for Orally
Inhaled Products. Springer, New York, pp. 65–66.
Suri, A., Chapel, S., Lu, C.,Venkatakrishnan, K., 2015. Physiologically based and population PK
modeling in optimizing drug development: a predict-learn-confirm analysis. Clin.
Pharmacol. Ther. 98(3), 336–344.
Terzano, C., 2008. Dry powder inhaler and the risk of error. Respiration.75(1), 14–15.
37
ACCEPTED MANUSCRIPT
Theil, F.P., Guentert, T.W., Haddad, S., Poulin, P., 2003. Utility of physiologically based
pharmacokinetic models to drug development and rational drug discovery candidate
selection. Toxicol. Lett. 138(1–2), 29–49.
Tong, Z.B., Zheng, B., Yang, R.Y., Yu, A.B., Chan H.K., 2013. CFD-DEM investigation of the
T
dispersion mechanisms in commercial dry powder inhalers. Powder Technol. 240, 19–24.
IP
Tong, Z.B., Kamiya, H., Yu, A.B., Chan H.K., Yang, R.Y., 2015. Multi-scale modelling of
CR
powder dispersion in a carrier-based inhalation system. Pharm. Res. 32(6), 2086–2096.
Versteeg, H.K., Hargrave, G., Harrington, L., Shrubb, I., Hodson, D., 2000. The use of
US
computational fluid dynamics (CFD) to predict pMDI air flows and aerosol plume
AN
formation. Respiratory Drug Delivery VII, 1, 257–264.
Voss, A., Finlay, W.H., 2002. Deagglomeration of dry powder pharmaceutical aerosols. Int. J.
M
Pharm. 248(1–2), 39–50.
ED
Weiss, P., Hersey, R.M., Dujovne, C.A., Bianchine, J.R., 1969. The metabolism of amiloride
hydrochloride in man. Clin. Pharmacol. Ther. 10(3), 401–406.
PT
Wong, W., Fletcher, D.F., Traini, D., Chan, HK., Crapper, J., Young, P.M., 2011. Particle
CE
Aerosolisation and Break-up in Dry Powder Inhalers: Evaluation and Modelling of the
Influence of Grid Structures for Agglomerated Systems. J. Pharm. Sci. 100(11), 4710-4721.
AC
Wong, W., Fletcher, D.F., Traini, D., Chan, HK., Young, P.M., 2012. The use of computational
approaches in inhaler development. Adv. Drug Deliv. Rev. 64(4), 312–322.
Wu, S., Salar-Behzadi, S., Fröhlich, E., 2013. Role of in-silico modelling in drug development
for inhalation treatment. J. Mol. Pharm. Org. Process Res. 1(2), 106.
38
ACCEPTED MANUSCRIPT
Wu, S., Zellnitz, S., Mercuri, A., Salar-Behzadi, S., Bresciani, M., Fröhlich, E., 2016. An in vitro
and in silico study of the impact of engineered surface modifications on drug detachment
from model carriers. Int. J. Pharm. 513(1–2), 109–117.
Zapletal, A., Desmond, K.J., Demizio, D., Coates, A.L., 1993. Lung recoil and the determination
T
of airflow limitation in cystic fibrosis and asthma. Pediatr. Pulmonol. 15(1), 13–18.
IP
Zhou, Q., Tong, Z., Tang, P., Yang, R., & Chan, H. K. 2013, June. CFD analysis of the
CR
aerosolization of carrier-based dry powder inhaler formulations. In A. Yu, K. Dong, R.
Yang, & S. Luding (Eds.), AIP Conference Proceedings, 1542(1), 1146–1149.
US
Zhou, D., Bui, K., Sostek, M., Al-Huniti, N., 2016. Simulation and prediction of the drug-drug
AN
interaction potential of naloxegol by physiologically based pharmacokinetic modeling. CPT
Pharmacometrics Syst. Pharmacol. 5(5), 250–257.
M
Zhu, C., Jiang, L., Chen, T.M., Hwang, K.K., 2002. A comparative study of artificial membrane
Chem. 37(5), 399–407.
ED
permeability assay for high throughput profiling of drug absorption potential. Eur. J. Med.
PT
Zhuang, X., Lu, C., 2016. PBPK modeling and simulation in drug research and development.
AC
CE
Acta Pharmaceutica Sinica B 6(5), 430–440.
39
ACCEPTED MANUSCRIPT
Fig. 1. Created model – solid domain: mouthpiece with grid area (a) and base with capsule storage (b)
Fig. 2. Created three-dimensional – fluid domain
Fig. 3. Meshed model
CR
Fig. 5. Velocity distribution detail (JM2), grid area – volume rendering
IP
T
Fig. 4. Velocity contours for JM2: velocity magnitude (a), tangential velocity (b), radial velocity (c)
Fig. 6. Outlet velocity vectors for jet-milled and spray dried particles
US
Fig. 7. Velocity streamlines (JM2) – isometric view
AN
Fig. 8. Particle tracking from the beginning of the simulation to the outlet
M
Fig. 9. Comparison of aerodynamic particle properties: ED (a), MMAD (b), and GSD (c) derived from CFD-
ED
DPM output data and ACI results taken from Djokic et al. (2014)
Fig. 10. Fractional and cumulative primary PSDs (geometric diameter) resulting from CFD-DPM
PT
simulations for five model formulations; values on x-scale denote the upper cut-off diameter
CE
Fig. 11. Regional particle deposition in the Aerolizer. Q=28.3 l/min
AC
Fig. 12. Predicted amiloride plasma concentration-time profiles, along with the mean data observed in
the in vivo studies for 10 mg oral solution (Jones et al., 1997) and 20 mg oral capsule (Weiss et al., 1969)
Fig. 13. Predicted amiloride plasma concentration-time profiles (M1-M4), along with the mean data
observed in the in vivo study for 1.91 mg dose from the nebulized solution (Jones et al., 1997); M1 refers
to the model using default ka(p), M2 refers to model using 25% reduced Ka(p), M3 refers to model using
50% reduced Ka(p), M4 refers to model using 75% reduced Ka(p) in comparison to the default value; inset
shows enlarged initial parts of the plasma concentration-time curves
40
ACCEPTED MANUSCRIPT
Fig. 14. Predicted regional absorption profile of amiloride following inhalation of a 1.91 mg dose from
the nebulized solution
Fig. 15. Simulated lung deposition pattern of the inhaled amiloride particles from different formulations
T
expressed as percent of delivered dose
CR
inset shows percent of delivered dose absorbed through the lungs
IP
Fig. 16. Simulated drug plasma concentration-time profiles for different amiloride DPI formulations;
Fig. 17. The dependence of regional particle deposition (expressed as percent of delivered dose) on the
AC
CE
PT
ED
M
AN
US
variations in airflow rate.
41
AC
CE
PT
ED
M
AN
US
CR
IP
T
ACCEPTED MANUSCRIPT
Fig. 1a
42
AC
Fig. 1b
CE
PT
ED
M
AN
US
CR
IP
T
ACCEPTED MANUSCRIPT
43
US
CR
IP
T
ACCEPTED MANUSCRIPT
AC
CE
PT
ED
M
AN
Fig. 2
44
AC
CE
PT
ED
M
AN
US
CR
IP
T
ACCEPTED MANUSCRIPT
Fig. 3
45
US
CR
IP
T
ACCEPTED MANUSCRIPT
AC
CE
PT
ED
M
AN
Fig. 4
46
ED
M
AN
US
CR
IP
T
ACCEPTED MANUSCRIPT
AC
CE
PT
Fig. 5
47
AC
CE
PT
ED
M
AN
US
CR
IP
T
ACCEPTED MANUSCRIPT
Fig. 6
48
AC
Fig. 7
CE
PT
ED
M
AN
US
CR
IP
T
ACCEPTED MANUSCRIPT
49
US
CR
IP
T
ACCEPTED MANUSCRIPT
AC
CE
PT
ED
M
AN
Fig. 8
50
AN
US
CR
IP
T
ACCEPTED MANUSCRIPT
AC
CE
PT
ED
M
Fig. 9a
51
AN
US
CR
IP
T
ACCEPTED MANUSCRIPT
AC
CE
PT
ED
M
Fig. 9b
52
AN
US
CR
IP
T
ACCEPTED MANUSCRIPT
AC
CE
PT
ED
M
Fig. 9c
53
AN
US
CR
IP
T
ACCEPTED MANUSCRIPT
AC
CE
PT
ED
M
Fig. 10
54
M
AN
US
CR
IP
T
ACCEPTED MANUSCRIPT
AC
CE
PT
ED
Fig. 11
55
AN
US
CR
IP
T
ACCEPTED MANUSCRIPT
AC
CE
PT
ED
M
Fig. 12
56
AN
US
CR
IP
T
ACCEPTED MANUSCRIPT
AC
CE
PT
ED
M
Fig. 13
57
M
AN
US
CR
IP
T
ACCEPTED MANUSCRIPT
AC
CE
PT
ED
Fig. 14
58
AN
US
CR
IP
T
ACCEPTED MANUSCRIPT
AC
CE
PT
ED
M
Fig. 15
59
AN
US
CR
IP
T
ACCEPTED MANUSCRIPT
AC
CE
PT
ED
M
Fig. 16
60
AN
US
CR
IP
T
ACCEPTED MANUSCRIPT
AC
CE
PT
ED
M
Fig. 17
61
ACCEPTED MANUSCRIPT
Table 1. Physicochemical and aerodynamic properties of amiloride HCl DPI samplesa
True density
EDc,e (%; mg)
FPFd,e (%)
(m)
(g/cm3)
JM1
0.84; 1.71; 4.01
1.67
68.15; 13.63
39.78
JM2
0.80; 2.07; 4.92
1.68
70.54; 14.11
41.25
JM3
0.87; 2.62; 6.58
1.72
82.97; 16.59
JM4
0.86; 2.96; 7.31
1.73
84.23; 16.85
SD
0.85; 2.89; 7.16
1.71
83.74; 16.75
GSDe
(m)
7.30
2.74
2.37
45.67
5.53
1.68
50.62
5.71
1.51
5.61
1.58
US
CR
IP
7.27
50.19
Taken from (Djokic et al., 2014); bintercepts (geometric particle size) for 10%, 50% and 90% of
AN
a
MMADe
T
Sample d10; d50; d90b
the cumulative particle mass; cED – emitted dose (percentage/mass of particles delivered from
M
the inhaler); dFPF – fine particle fraction (percentage of the mass of drug particles that have
ED
aerodynamic diameter of less than 5 m), eThese values were not used as inputs for the
AC
CE
PT
simulations
62
ACCEPTED MANUSCRIPT
Table 2. Summary of input parameters for PBPK simulations
Value
Molecular weight (g/mol)
229.63
logD (pH 7.4)
-0.86a
pKa
8.7b
Solubility (aq) (mg/ml)
5.2b
IP
T
Parameter
CR
Human jejunal permeability, Peff (x 10-4 cm/s)
US
Diffusion coefficient (x 10-6 cm2/s)
Fluid intake with oral dose (ml)
AN
Mean precipitation time (s)
Drug particle density (g/ml)
ED
MMAD for the nebulized solution (m)
M
Particle radius for oral dosing (µm)
1.6c
8.86d
250e
900e
25e
4f
1.2e
1e
Blood/plasma concentration ratio
60g
Body weight (kg)
70
CE
PT
Unbound percent in plasma (%)
25h
Unbound percent in enterocytes (%)
5h
a
AC
Renal clearance, CLren (l/h)
Taken from (Zhu et al., 2002); bTaken from (INCHEM, 1997); cTaken from (Dahlgren et al.,
2015); dGastroPlus™ predicted; eGastroPlus™ default values; fTaken from (Jones et al., 1997);
g
Taken from (Spahn et al., 1987); hOptimized values
63
ACCEPTED MANUSCRIPT
Table 3. Comparison of the predicted and observed pharmacokinetic parameters for 10 mg and
tmax (h)
AUC0-∞ (ng h/ml)
20.35 (-11.20)
2.44 (18.39)
380.80 (-17.66)
In vivo meana
18.30
2.99
323.64
In vivo rangeb
9.80-30.60
1.90-4.40
160.00-390.00
Predicted (%PE)
40.70 (-22.59)
2.44 (7.22)
761.58 (-0.59)
In vivo meanc
33.20
2.63
757.14
In vivo ranged
28.80-61.30
/
/
1.91 mg nebulized
Predicted M1 (%PE)
1.45 (-9.02)
0.32 (38.46)
25.79 (-46.45)
solution
Predicted M2 (%PE)
1.44 (-8.27)
0.36 (30.77)
25.79 (-46.45)
1.35 (-1.50)
0.48 (7.69)
25.79 (-46.45)
Predicted M4 (%PE)
1.21 (9.02)
0.9 (-73.08)
25.79 (-46.45)
In vivo meana
1.33
0.52
17.61
In vivo rangee
0.10-3.20
0.30-0.70
3.20-32.00
Refer to the mean plasma concentration–time profile from a single study (Jones et al., 1997);
b
CE
a
PT
ED
M
Predicted M3 (%PE)
IP
20 mg IR capsule
CR
10 mg oral solution Predicted (%PE)
US
Parameter
T
Cmax (ng/ml)
AN
20 mg amiloride oral dose, and 1.91 mg delivered dose from a nebulized solution
Refer to the range of values from different studies (Jones et al., 1997; Spahn et al., 1987); cRefer
AC
to the mean plasma concentration–time profile from a single study (Weiss et al., 1969); dRefer to
the range of values from different studies (Weiss et al., 1969; Grayson et al., 1971); eRefer to the
range of values from a single study (Jones et al., 1997)
64
ED
M
AN
US
CR
IP
T
ACCEPTED MANUSCRIPT
AC
CE
PT
Graphical abstract
65
Документ
Категория
Без категории
Просмотров
1
Размер файла
1 810 Кб
Теги
022, 2017, ejps
1/--страниц
Пожаловаться на содержимое документа